S390 [liftoff]: Implement simd i32 convert low to fp

Change-Id: I6c24b0930d489a9af7e877e783663f2493168604
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3417920Reviewed-by: Junliang Yan <junyan@redhat.com>
Commit-Queue: Milad Farazmand <mfarazma@redhat.com>
Cr-Commit-Position: refs/heads/main@{#78793}

src/codegen/s390/macro-assembler-s390.cc

@@ -5537,6 +5537,18 @@ void TurboAssembler::I16x8BitMask(Register dst, Simd128Register src,
   vlgv(dst, scratch2, MemOperand(r0, 7), Condition(0));
 }
 
+void TurboAssembler::F64x2ConvertLowI32x4S(Simd128Register dst,
+                                           Simd128Register src) {
+  vupl(dst, src, Condition(0), Condition(0), Condition(2));
+  vcdg(dst, dst, Condition(4), Condition(0), Condition(3));
+}
+
+void TurboAssembler::F64x2ConvertLowI32x4U(Simd128Register dst,
+                                           Simd128Register src) {
+  vupll(dst, src, Condition(0), Condition(0), Condition(2));
+  vcdlg(dst, dst, Condition(4), Condition(0), Condition(3));
+}
+
 void TurboAssembler::I8x16BitMask(Register dst, Simd128Register src,
                                   Register scratch1, Register scratch2,
                                   Simd128Register scratch3) {

src/codegen/s390/macro-assembler-s390.h

@@ -1114,29 +1114,31 @@ class V8_EXPORT_PRIVATE TurboAssembler : public TurboAssemblerBase {
   void I8x16BitMask(Register dst, Simd128Register src, Register scratch1,
                     Register scratch2, Simd128Register scratch3);
 
-#define SIMD_UNOP_LIST(V) \
-  V(F64x2Abs)             \
-  V(F64x2Neg)             \
-  V(F64x2Sqrt)            \
-  V(F64x2Ceil)            \
-  V(F64x2Floor)           \
-  V(F64x2Trunc)           \
-  V(F64x2NearestInt)      \
-  V(F32x4Abs)             \
-  V(F32x4Neg)             \
-  V(F32x4Sqrt)            \
-  V(F32x4Ceil)            \
-  V(F32x4Floor)           \
-  V(F32x4Trunc)           \
-  V(F32x4NearestInt)      \
-  V(I64x2Abs)             \
-  V(I32x4Abs)             \
-  V(I16x8Abs)             \
-  V(I8x16Abs)             \
-  V(I64x2Neg)             \
-  V(I32x4Neg)             \
-  V(I16x8Neg)             \
-  V(I8x16Neg)             \
+#define SIMD_UNOP_LIST(V)  \
+  V(F64x2Abs)              \
+  V(F64x2Neg)              \
+  V(F64x2Sqrt)             \
+  V(F64x2Ceil)             \
+  V(F64x2Floor)            \
+  V(F64x2Trunc)            \
+  V(F64x2NearestInt)       \
+  V(F64x2ConvertLowI32x4S) \
+  V(F64x2ConvertLowI32x4U) \
+  V(F32x4Abs)              \
+  V(F32x4Neg)              \
+  V(F32x4Sqrt)             \
+  V(F32x4Ceil)             \
+  V(F32x4Floor)            \
+  V(F32x4Trunc)            \
+  V(F32x4NearestInt)       \
+  V(I64x2Abs)              \
+  V(I32x4Abs)              \
+  V(I16x8Abs)              \
+  V(I8x16Abs)              \
+  V(I64x2Neg)              \
+  V(I32x4Neg)              \
+  V(I16x8Neg)              \
+  V(I8x16Neg)              \
   V(S128Not)
 
 #define PROTOTYPE_SIMD_UNOP(name) \

src/compiler/backend/s390/code-generator-s390.cc

@@ -3257,17 +3257,13 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       break;
     }
     case kS390_F64x2ConvertLowI32x4S: {
-      __ vupl(kScratchDoubleReg, i.InputSimd128Register(0), Condition(0),
-              Condition(0), Condition(2));
-      __ vcdg(i.OutputSimd128Register(), kScratchDoubleReg, Condition(4),
-              Condition(0), Condition(3));
+      __ F64x2ConvertLowI32x4S(i.OutputSimd128Register(),
+                               i.InputSimd128Register(0));
       break;
     }
     case kS390_F64x2ConvertLowI32x4U: {
-      __ vupll(kScratchDoubleReg, i.InputSimd128Register(0), Condition(0),
-               Condition(0), Condition(2));
-      __ vcdlg(i.OutputSimd128Register(), kScratchDoubleReg, Condition(4),
-               Condition(0), Condition(3));
+      __ F64x2ConvertLowI32x4U(i.OutputSimd128Register(),
+                               i.InputSimd128Register(0));
       break;
     }
     case kS390_F64x2PromoteLowF32x4: {

src/wasm/baseline/s390/liftoff-assembler-s390.h

@@ -2515,12 +2515,12 @@ void LiftoffAssembler::emit_i8x16_swizzle(LiftoffRegister dst,
 
 void LiftoffAssembler::emit_f64x2_convert_low_i32x4_s(LiftoffRegister dst,
                                                       LiftoffRegister src) {
-  bailout(kSimd, "f64x2.convert_low_i32x4_s");
+  F64x2ConvertLowI32x4S(dst.fp(), src.fp());
 }
 
 void LiftoffAssembler::emit_f64x2_convert_low_i32x4_u(LiftoffRegister dst,
                                                       LiftoffRegister src) {
-  bailout(kSimd, "f64x2.convert_low_i32x4_u");
+  F64x2ConvertLowI32x4U(dst.fp(), src.fp());
 }
 
 void LiftoffAssembler::emit_f64x2_promote_low_f32x4(LiftoffRegister dst,